Assessing Network Middlebox Impact on End-to-End Protocol Behavior via a Distributed and Reprogrammable Framework
This dissertation focuses on the detection and verification of network middleboxes thanks to the creation of a new distributed framework called NoPASARAN.
Overview
Network middleboxes are intermediate devices that violate the end-to-end principle upon which the Internet was originally designed. While these devices can serve legitimate purposes, such as firewalls protecting intranets or caches reducing latency by serving static content, they can also be used for more questionable or malicious activities. Examples include transparent proxies employed for website censorship or malicious nodes carrying out so-called Man-in-the-Middle (MITM) attacks. This dissertation focuses on the detection and verification of network middleboxes thanks to the creation of a new distributed framework called NoPASARAN. We demonstrate how this framework enhances protocol compliance, particularly for widely-used protocols like the Hypertext Transfer Protocol (HTTP), and aids in uncovering middlebox behaviors and policy vulnerabilities.
Presenters
Brief Biography
Ilies Benhabbour is a senior Ph.D. student with an international academic and research background spanning France, Germany, and Saudi Arabia. He earned his bachelor's degree from UVSQ and a master's degree from Eurecom. His early work explored adversarial attacks in machine learning and explainable AI at iABG in Germany. He is currently pursuing his Ph.D. under the supervision of Prof. Marc Dacier, focusing on network middlebox interference.
